Objective: The aim of this study was to evaluate the classification power as the age and sex of discriminant models (DM) generated from the echo-texture of the Achilles tendon (AT). Materials and Methods: The design was an observational, cross-sectional study report on 103 subjects (49 females and 54 males) 18 to 53 years old (mean = 29.8, SD = 0.86). We took bilateral transverse ultrasonographies at 20 mm from the insertion of the AT. Sixty variables were obtained from grey level co-occurrence matrices (GLCM): contrast, homogeneity, uniformity, entropy and correlation for three scales (1, 5 and 10 pixels) and two orientations (0 º and 90 º) with ImageJ 1.43 software. Multidimensional discriminant models were performed with forced inclusion method (C.I. 95%). Results: The global MD for classification by age group showed a sensitivity of 74% and a specificity of 91%. In the DM women’s ultrasound pattern sensitivity increased to 90% and specificity 97%, while sensitivity and specificity of ultrasound pattern DM men were 96% and 99% respectively. When cross-validations were performed for models of women and men, the sensitivities of the models were reduced to 30%, suggesting that the AT echotextural behavior is different. Conclusion: Multidimensional analysis of Achilles tendon echotexture with GLCM showed good sensitivity and specificity in the classification of subjects by age group that improved when the models were segmented by gender.
Echo-texture, Grey Level Co-occurrence Matrices, Quantitative Ultrasonography, Tendon; Discriminant Analysis.
Presenting author: JOSÉ RÍOS-DIAZ
Organisation: Group Research “Preventive Ultrasonography &Morphodensitometry and Physiotherapy and Manual Therapy Research”, Catholic University of San Antonio, Guadalupe, Murcia, Spain
co-authors: Jacinto J. Martínez-Payá; Ana de-Groot-Ferrando; María Elena del-Baño-Aledo